Geophysical Research Letters

Characterization of moderate ash-and-gas explosionsat Santiaguito volcano, Guatemala, from infrasoundwaveform inversion and thermalinfrared measurements

S. De Angelis1, O. D. Lamb1, A. Lamur1, A. J. Hornby1, F. W. von Aulock1, G. Chigna2, Y. Lavallée1,and A. Rietbrock1

1School of Ocean and Environmental Sciences, University of Liverpool, Liverpool, UK, 2Instituto Nacional de Sismología,Vulcanología, Meteorología e Hidrología (INSIVUMEH), Guatemala City, Guatemala

Abstract The rapid discharge of gas and rock fragments during volcanic eruptions generates acousticinfrasound. Here we present results from the inversion of infrasound signals associated with small andmoderate gas-and-ash explosions at Santiaguito volcano, Guatemala, to retrieve the time history of masseruption rate at the vent. Acoustic waveform inversion is complemented by analyses of thermal infraredimagery to constrain the volume and rise dynamics of the eruption plume. Finally, we combine resultsfrom the two methods in order to assess the bulk density of the erupted mixture, constrain the timing ofthe transition from a momentum-driven jet to a buoyant plume, and to evaluate the relative volumefractions of ash and gas during the initial thrust phase. Our results demonstrate that eruptive plumesassociated with small-to-moderate size explosions at Santiaguito only carry minor fractions of ash,suggesting that these events may not involve extensive magma fragmentation in the conduit.

1. Introduction

Volcanoes are efficient sources of infrasonic acoustic waves with frequencies below 20 Hz [Johnson et al.,2004; Fee and Matoza, 2013; Garcés et al., 2013]. The sudden release of hot gases and pyroclasts from volcanicvents generates infrasound via volumetric acceleration of the atmosphere. Over short source-receiver dis-tances, acoustic wave propagation in the atmosphere is characterized by a relatively simple Green’s function.Therefore, at local distances from the source (<5 km) the effects of scattering have frequently been neglectedand infrasound used to investigate the dynamics of volcanic explosions and their source mechanisms[Kim et al., 2012]. Acoustic sources with dimensions significantly smaller than the characteristic wavelengthof infrasound can be considered acoustically compact and, thus, approximated using multipole expansion[Rossing, 2016]. Several authors used infrasound to assess the strength of volcanic explosions measuring eithergas velocity or volume flux at the vent. Johnson [2007] and Johnson and Miller [2014] described explosionsat Karymsky and Sakurajima volcanoes as monopoles, which is an isotropically expanding source causingvolumetric acceleration of the atmosphere and radiating infrasound uniformly in all directions; Delle Donneand Ripepe [2012] demonstrated the use of a monopole source model to assess the volumetric flux duringexplosions at Stromboli volcano. A dipole source model (i.e., two monopoles of equal strength and oppositephase, separated by a small distance) was used to evaluate gas velocity at the vent for eruptions at Shishaldin[Vergniolle and Caplan-Auerbach, 2004], Augustine [Caplan-Auerbach et al., 2010], Eyjafjallajökull [Ripepe et al.,2013], and Redoubt [Lamb et al., 2015] volcanoes. Matoza et al. [2009] argued that the infrasound wavefieldproduced by sustained eruptions is best described by jet noise models, and thus, an acoustic quadrupole mayrepresent a more accurate representation of the radiation mechanisms at the source. Kim et al. [2012] recentlyproposed a waveform inversion workflow based on a representation of the acoustic source as a combinationof multipole terms, using analytical Greens functions in a half-space. This method can simultaneously retrievethe time history of mass outflow at the vent and account for directivity effects on the acoustic wavefield. Inthis study, we applied the waveform inversion method of Kim et al. [2012] to infrasound signals from smalland moderate gas-and-ash explosions at the Santiaguito lava dome complex, Guatemala. For one explosion,we complemented infrasound with the analysis of thermal infrared data and measurements of ash density toconstrain the relative volume fractions of gas and ash in the eruptive plume.

RESEARCH LETTER10.1002/2016GL069098

Key Points:• Infrasound waveform inversion allows

assessing source parameters ofvolcanic explosions

• Thermal infrared imagery allowsinvestigating the transition from aninitial gas thrust phase to a buoyantplume during volcanic eruptions

• Eruption plumes produced bysmall-to-moderate eruptions atSantiaguito are gas rich

Supporting Information:• Supporting Information S1

Correspondence to:S. De Angelis,silvioda@liverpool.ac.uk

Citation:De Angelis, S., O. D. Lamb, A. Lamur,A. J. Hornby, F. W. von Aulock,G. Chigna, Y. Lavallée, and A. Rietbrock(2016), Characterization of moderateash-and-gas explosions at Santiaguitovolcano, Guatemala, from infrasoundwaveform inversion and thermalinfrared measurements, Geophys.Res. Lett., 43, 6220–6227,doi:10.1002/2016GL069098.

Received 11 APR 2016

Accepted 28 MAY 2016

Accepted article online 7 JUN 2016

Published online 27 JUN 2016

©2016. American Geophysical Union.All Rights Reserved.

DE ANGELIS ET AL. EXPLOSIONS AT SANTIAGUITO 6220

Geophysical Research Letters 10.1002/2016GL069098

Figure 1. (a) Map showing the infrasound network and thermal infrared observation point at Santiaguito. (b) Unfiltered,individually normalized, infrasound waveforms recorded across the network showing a propagation velocity of 325 m/s.

2. Activity at Santiaguito

Santiaguito is a complex of dacitic lava domes that formed within the collapse scar of the 1902 eruption ofSanta Maria volcano, one of the largest in the twentieth century [Bennett et al., 1992]. The ongoing eruptionof Santiaguito began in 1922 [Bluth and Rose, 2004], with present activity concentrated at the eastern Calientevent (Figure 1a). Eruptive activity consists of continuous extrusion of blocky lava flows interspersed byfrequent gas-and-ash explosions, accompanied by occasional collapses and short runout pyroclastic flows.Explosions throughout the past decade have been of small to moderate size and have produced volatile-richplumes with relatively minor proportions of ash [Yamamoto et al., 2008]. Recent petrological and geophysicalanalyses suggest that these small gas-and-ash explosions are not triggered by typical magmatic fragmenta-tion but rather through sacrificial fragmentation that occurs along fault zones due to shear-induced thermalvesiculation [Lavallée et al., 2015].

3. Data and Methods

In November 2014 we deployed a network of four iTem prs100 [Delle Donne and Ripepe, 2012] infrasound sen-sors (flat response between 0.01 and 100 Hz, sensitivity 25 mV/Pa,±100 Pa full scale, and self-noise−52 dB rel.1 Pa) at Santiaguito, all within 5 km of the active Caliente vent (Figure 1a). During the initial deployment wealso gathered thermal infrared (TIR) imagery of the explosive activity from the summit of Santa Maria, whichoverlooks the Caliente vent providing a unique observation point at Santiaguito (Figure 1a). The infrasoundwaveforms recorded during November–December 2014 were characterized by relatively low amplitudes(0.7–2 Pa at a distance of 4.5 km from the vent), had durations of between 5 and 10 s, and included either oneor multiple pulses. Unlike previous observations by Johnson and Lees [2010], we recorded infrasound signalslacking significant coda phases, and waveform coherency across the network was high (Figure 1b). For thisstudy we used acoustic signals from 61 explosions during the period 29 November to 5 December 2014,when all infrasound sensors were operational. We also analyzed TIR imagery gathered during an explosionon 30 November 2014, with a FLIR T450sc infrared camera equipped with a 30 mm lens (FOV: 15× 11.25,IFOV: 0.82 mrad) calibrated in the relevant temperature range, and considering atmospheric conditions atSantiaguito.

3.1. Acoustic Multipole Source InversionWe applied the acoustic multipole source inversion method of Kim et al. [2012] in order to assess the timehistory of mass eruption rate at the vent (monopole source term) and the azimuth and strength of a hori-zontal force (dipole source term) to represent directivity effects within the crater region of the Caliente vent.Following the formalism of Kim et al. [2012], a discrete time series of acoustic pressures radiated by a sourceconsisting of a monopole and a horizontal dipole can be written as

pki = 1

2𝜋ri

[mk

1 +xi

crimk

2 +yi

crimk

3

](1)

DE ANGELIS ET AL. EXPLOSIONS AT SANTIAGUITO 6221

Geophysical Research Letters 10.1002/2016GL069098

where pki is the kth element of the time series recorded at the ith station, c is the atmospheric sound speed,

and ri is the distance between the source and the ith receiver. The vector mk is

mk =[

mk1,mk

2,mk3

]=[

S(t0 + kΔt), Fx(t0 + kΔt), Fy(t0 + kΔt)]

(2)

where S is the strength of the monopole source term (i.e., the mass eruption rate), measured in kg/s, andFx and Fy are the horizontal components of the dipole force, measured in kg/ms2. The system of linearequations in (1) can be rewritten in matrix form as

Pk = Gmk (3)

where Pk is a vector of sampled acoustic pressures at each station. Inversion of equation (3) for mk allowsassessing monopole and dipole strengths with the only additional requirement that the number of stationsmust be greater than three. Kim et al. [2012] suggested, based on synthetic tests, that the best results for aninversion are obtained from networks of at least three sensors evenly distributed around the source (i.e., withan azimuthal gap of 120∘). Our network at Santiaguito satisfies these requirements reasonably, with anazimuthal gap of about 159∘. This is the smallest gap that could be achieved given restrictions to access inthe western and southwestern sectors of the volcano (Figure 1a).

For the inversion, only high signal-to-noise ratio waveforms associated with confirmed explosions and atmo-spheric propagation of a plume were selected. The inversion was performed over 5 s signal windows includingthe onset of each event. Linear piecewise detrending was applied to all signals according to the Finite WindowZero Pressure Zero Flux correction proposed by Johnson and Miller [2014] in order to mitigate effects of instru-ment drift. Waveforms were filtered between 0.1 and 4 Hz (2-pole, Butterworth filter) to include the dominantpart of the energy spectrum while mitigating the effects of high-frequency noise on the stability of theinversion. We used an atmospheric sound velocity of 325 m/s as measured from the moveout of infrasoundarrivals across the network (Figure 1b).

3.2. Analysis of Thermal Infrared ImageryTIR video footage was analyzed following the workflow of Valade et al. [2014], which allows tracking of ascend-ing volcanic plumes. Thermal images were corrected and transformed from units of pixels to metric units,considering the focal length of the lens, the vertical field of view, the instantaneous field of view, and thedistance and inclination between vent and the observation point (the reader may refer to Valade et al. [2014]for additional details on these procedures). Following transformation of all TIR video frames to a vent-centeredsystem of coordinates, and assuming symmetrical plume growth about a vertical axis, a number of eruptionparameters were measured, including plume ascent velocity and volume throughout the duration of theeruptive event.

4. Results

We performed multipole source inversion for 61 explosions recorded at Santiaguito during 29 November to5 December 2014. Results from waveform inversions are illustrated in Figure 2 for an explosion recorded on30 November at 01:09:14 UTC. The inversion produced a good fit between synthetic signals and observations(Figure 2d). As a measure of fit quality, we calculated variance reduction for the ensemble of N waveforms:

VR = 1N

N∑i=1

[1 −

∫ (di − si)2

∫ (di)2

]⋅ 100 (4)

where si and di are the ith synthetic and data, respectively, and the integrals are performed over the durationof the signals. For the explosion in Figure 2 we obtained VR = 92.3%. The most obvious, yet minor, mismatchbetween data and synthetics is seen at stations LB01 and LB03 (Figure 2d), later in the waveform, past thewell-fit initial pulse. This was consistently observed for all 61 events analyzed, likely due to the location ofstations LB01 and LB03 at comparatively larger distances from the vent. It has been shown that propagation ofthe acoustic wavefield over terrain with variable topography may result in diffraction, attenuation, or focusingeffects beyond the predictions of geometrical spreading [Lacanna et al., 2014]. Recent studies have beguntackling the influence of variable topography and atmospheric conditions on the acoustic wavefield [Lacannaet al., 2014; Kim et al., 2015] suggesting that numerical Green’s functions hold great promise for integrating

DE ANGELIS ET AL. EXPLOSIONS AT SANTIAGUITO 6222

Geophysical Research Letters 10.1002/2016GL069098

Figure 2. Acoustic multipole source inversion for an explosion at Santiaguito on 30 November 2014 01:09:14 UTC:(a) dipole azimuth over the source time function, (b) dipole strength, (c) monopole strength, and (d) fit betweensynthetic and observed waveforms.

these complex phenomena in infrasound propagation studies and waveform inversion. Here, however, thegood agreement between synthetics and data confirms that the effects of topography, where present, arecomparatively small. This can be attributed to the location of the sensors that, with the only exception of LB03,benefit from unobstructed source-receiver paths (supporting information Figure 1S). The dominant directionof the dipole azimuth for this event is 120∘ (Figure 2a), suggesting ESE directivity. Figure 3 illustrates the resultsof multipole source inversions for all 61 events analyzed; peak monopole and dipole strength, and dipoleazimuth, are stable throughout the study period, reflecting the regular character of activity at Santiaguito. Inorder to assess the stability of our results, and the sensitivity of the method to changes in network geometry,we conducted a test performing waveform inversion for all explosions using all possible combinations of onlythree stations. Results are illustrated by the confidence bars in Figure 3, which represent the range of valuesobtained from all inversions for each of the parameters evaluated. It is worth noting that the largest deviationfrom the results using data from the entire network was consistently observed removing station LB01 fromthe inversion. This was expected as removal of station LB01 introduces the largest azimuthal gap among allpossible three-station network configurations.

Results of the analysis of TIR data recorded on 30 November, for the explosion at 01:09:14 UTC, are illustratedin Figure 4. Figures 4a and 4b display infrasound recorded at station LB05, the time history of plume volume,and its ascent velocity, respectively; Figure 4c illustrates the temporal evolution of the cross section (side view)of the rising plume. The time series of plume ascent velocity (measured at the plume front) allows identifyingthe transition from an early, short-duration, thrust phase dominated by a momentum-driven volcanic jet, toa buoyant regime where the morphology of the plume was controlled by entrapment of surrounding, colder,atmospheric air [Patrick, 2007; Marchetti et al., 2009; Delle Donne and Ripepe, 2012]. The timing of this tran-sition, which corresponds to the point when the plume ascent velocity stabilizes around values <10 m/s inFigure 4c (point 1 at about 4.2 s), can be used to constrain the cumulative volume erupted before engulfmentof atmospheric air became dominant. From the data presented in Figure 4b, we estimated this volume to be2.2 ⋅ 105 m3. We suggest that this value provides a reasonable first-order approximation for the cumulativevolume of ash and gas injected from the volcanic vent into the atmosphere during the explosion. Finally, themonopole strength function (Figure 2c) was integrated over the duration of the thrust phase to provide anestimate of the cumulative erupted mass of 5.9 ⋅ 104 kg. Taken together, these values suggest a bulk densityof the plume, during the initial thrust phase, of 0.27 kg/m3.

5. Discussion and Conclusive Remarks

An integrated, multidisciplinary, approach can help to improve our understanding of the mechanisms of vol-canic explosions and the dynamics of plume rise. Here we have shown that the strength of acoustic sources,

DE ANGELIS ET AL. EXPLOSIONS AT SANTIAGUITO 6223

Geophysical Research Letters 10.1002/2016GL069098

Figure 3. Results of acoustic multipole source inversion for 61 explosions recorded at Santiaguito, between 29November and 5 December 2014. (a) peak monopole strength, (b) peak dipole strength, and (c) dipole azimuth.Confidence bars reflect the range of values obtained from inversions for all possible combinations of three stationsin the network (see text in the manuscript). The dashed line represents the median value of measurements.

and the directivity of the associated wavefield, during volcanic explosions can be estimated by inversion ofinfrasound waveforms, and the erupted volume can be retrieved from analysis of TIR or visual imagery. Thestrength of the acoustic monopole source time function presented here, in Figure 2c, is consistent with pre-vious measurements of eruption rates at Santiaguito [Cerminara et al., 2015] based on plume modeling andinversion of TIR data. The strength of the dipole function at Santiaguito (Figure 2b) is about 2 orders of mag-nitude smaller than values reported for larger Vulcanian explosions at Tunghuraua volcano [Kim et al., 2012]and Strombolian explosions at Mount Erebus [Johnson et al., 2008]. However, we had anticipated a compar-atively weaker dipole as peak infrasound amplitudes at Santiaguito are 2 orders of magnitude smaller thanthose recorded at Tunghuraua and Mount Erebus at similar distances from the vent. We stress that this isnot equivalent to a weak (negligible) dipole in relation to the strength of the monopole source term. On thecontrary, the dipole term plays a crucial role in accounting for the directivity of infrasound although the originof the directional nature the acoustic wavefield in the source region cannot be fully resolved. In the future,numerical modeling needs to be integrated within the acoustic inversion workflow to evaluate the influenceof factors such as near-source topography (e.g., the geometry of the crater wall) on volcanic infrasound. Inorder to confirm the importance of the dipole term for our results, we attempted to estimate mass eruptionrate via direct integration of the infrasound time series (i.e., a monopole-only source) recorded at each stationin the network following Johnson and Miller [2014]. We observed significant variability between stations andcould not obtain stable results, with LB03 showing the largest deviation from the multipole inversion.

Acoustic multipole source inversion relies on the assumption that the source can be considered acousticallycompact [Rossing, 2016]; that is, its largest dimension is small relative to the dominant wavelength of thesignal. Previous studies suggested that infrasound at Santiaguito may be generated by uplift of the entiredome (a roughly circular structure, about 200 m in diameter), before the onset of explosions, a source thatwould be inherently noncompact [Johnson and Lees, 2010]. Our recent observations show that the initial

DE ANGELIS ET AL. EXPLOSIONS AT SANTIAGUITO 6224

Geophysical Research Letters 10.1002/2016GL069098

Figure 4. Time synchronized infrasound and thermal infrared (TIR) data for the explosion in Figure 2. (a) Infrasoundsignal recorded at station LB05, (b) plume volume and plume velocity from the analysis of TIR imagery, and(c) cross-sectional view of the plume from the analysis of TIR. Frames 1–3 correspond to times 1–3 in Figure 4b.

compression in the infrasound signals coincides with the onset of the plume in the visual and TIR images(considering infrasound propagating at 325 m/s and GPS time synchronization within 1/30 s between TIR andinfrasound). Video footage from recent explosions at Santiaguito also revealed that they may initiate at sub-parallel linear faults near the center of the dome [von Aulock et al., 2016], rather than along arcuate or ringfaults along its periphery as it was commonplace in the past. The typical dimension of these linear fissures isof the order of 20–30 m, and thus, they can be considered acoustically compact with respect to the dominantwavelength of infrasound at Santiaguito (350–650 m).

Analysis of TIR imagery allowed identifying the timing of the transition between an early, short-duration,gas thrust phase and a buoyant plume. We measured both the cumulative erupted volume and mass at thispoint, which we consider reliable proxies for the total amount of material injected from the vent into theatmosphere. They possibly represent a slight underestimate as ash and gas venting may continue, albeit withlower intensity, during the buoyant phase. Bluth and Rose [2004] had previously reported the absence of a gasthrust phase at Santiaguito and suggested that plume rise was controlled by buoyancy at the source. Theirinference was supported by the observation that past eruptive activity originated, predominantly, at annularcracks along the periphery of the dome. While our data confirm that late plume rise at Santiaguito is con-trolled by buoyancy, they also demonstrate that during recent explosions, an early thrust phase was present.This observation is consistent with eruptive plumes sourced at compact, elongated, fractures such as thoserecently identified at Santiaguito [von Aulock et al., 2016].

Finally, we estimated the bulk density of the initial plume, 𝜌plume= 0.27 kg/m3, at the time of the gas thrust-buoyancy transition. Additionally, we collected samples of volcanic ash and measured its density, 𝜌ash=2650 kg/m3, using He pycnometry. These data can be used to estimate the relative volume fractions of ashand gas erupted during the explosion. We considered that the main volatile phase in the plume is water vapor[Yamamoto et al., 2008] with density 𝜌gas = 0.15–0.21 kg/m3 (at eruption temperatures of 1000–1400 K and

DE ANGELIS ET AL. EXPLOSIONS AT SANTIAGUITO 6225

Geophysical Research Letters 10.1002/2016GL069098

decompression to atmospheric pressure). The volume fraction of ash in a plume 𝜙 can, then, be calculatedfrom the simple equation:

𝜙 =𝜌plume − 𝜌gas

𝜌ash − 𝜌gas(5)

For the range of plume, ash, and gas densities considered in our study, 𝜙= 2.3–4.5 ⋅ 10−5, which is about1 order of magnitude lower than previous, independent, estimates by Yamamoto et al. [2008]. We note thatthese results are sensitive to the particular choice of ash density. If ash density is considered to be 650 kg/m3,as in Yamamoto et al. [2008], our estimates of𝜙would fall in the range 9.2 ⋅10−5 –1.8 ⋅10−4, closer to the valuespresented by these authors.

In conclusion, we have demonstrated that combined analyses of infrasound, TIR, visual, and field data canshed light on the dynamics of explosions at lava dome volcanoes. We have investigated moderate gas-and-ashexplosions at the Santiaguito lava dome complex and observed that eruptive plumes exhibit transitionsbetween an early, short-duration, thrust phase and later buoyant rise. We assessed the relative volume frac-tions of gas and ash erupted during a single explosion and demonstrated that plumes from small-to-moderatesize events at Santiaguito are extremely gas rich. The minor volume fraction of ash in the plume suggests thatthese events may not involve significant fragmentation of magma in the conduit and concurs with the recentsuggestion that local sacrificial fragmentation along fault zones, due to shear-induced thermal vesiculation,may be at the origin of such events [Lavallée et al., 2015].

We surmise that the combined use of infrasound and TIR, or visual, imagery holds great promise for assess-ment of mass eruption rates and volcanic plume rise, with potential applications in near real time. We envisionthat numerical Green’s functions will be included in future developments of infrasound waveform inversionschemes in order to account for the effects of near-vent geometry, local topography, and variable atmosphericconditions on volcanic infrasound. For larger eruptions, mass eruption rate is a key input parameter in modelsof atmospheric plume rise and ash dispersal, and thus, its rapid assessment is crucial to mitigating the risksassociated with volcanic eruptions.

ReferencesBennett, E. H. S., F. M. Conway, and W. I. Rose (1992), Santa María, Guatemala: A decade volcano, Eos Trans. AGU, 73(48), 521–521,

doi:10.1029/91EO00387.Bluth, G. J., and W. I. Rose (2004), Observations of eruptive activity at Santiaguito volcano, Guatemala, J. Volcanol. Geotherm. Res., 136(3–4),

297–302, doi:10.1016/j.jvolgeores.2004.06.001.Caplan-Auerbach, J., A. Bellesiles, and J. K. Fernandes (2010), Estimates of eruption velocity and plume height from infrasonic recordings of

the 2006 eruption of Augustine Volcano, Alaska, J. Volcanol. Geotherm. Res., 189(1–2), 12–18, doi:10.1016/j.jvolgeores.2009.10.002.Cerminara, M., T. E. Ongaro, S. Valade, and A. J. Harris (2015), Volcanic plume vent conditions retrieved from infrared images: A forward and

inverse modeling approach, J. Volcanol. Geotherm. Res., 300, 129–147, doi:10.1016/j.jvolgeores.2014.12.015.Delle Donne, D., and M. Ripepe (2012), High-frame rate thermal imagery of Strombolian explosions: Implications for explosive and

infrasonic source dynamics, J. Geophys. Res., 117, B09206, doi:10.1029/2011JB008987.Fee, D., and R. S. Matoza (2013), An overview of volcano infrasound: From Hawaiian to plinian, local to global, J. Volcanol. Geotherm. Res.,

249, 123–139, doi:10.1016/j.jvolgeores.2012.09.002.Garcés, M. A., D. Fee, and R. Matoza (2013), Volcano acoustics, in Modeling Volcanic Processes: The Physics and Mathematics of Volcanism,

pp. 359–383, Cambridge Univ. Press.Johnson, J., R. Aster, K. R. Jones, P. Kyle, and B. McIntosh (2008), Acoustic source characterization of impulsive Strombolian eruptions from

the Mount Erebus lava lake, J. Volcanol. Geotherm. Res., 177(3), 673–686, doi:10.1016/j.jvolgeores.2008.06.028.Johnson, J. B. (2007), On the relation between infrasound, seismicity, and small pyroclastic explosions at Karymsky Volcano, J. Geophys. Res.,

112, B08203, doi:10.1029/2006JB004654.Johnson, J. B., and J. M. Lees (2010), Sound produced by the rapidly inflating Santiaguito Lava Dome, Guatemala, Geophys. Res. Lett., 37,

L22305, doi:10.1029/2010GL045217.Johnson, J. B., and A. J. C. Miller (2014), Application of the monopole source to quantify explosive flux during vulcanian explosions at

Sakurajima Volcano (Japan), Seismol. Res. Lett., 85(6), 1163–1176, doi:10.1785/0220140058.Johnson, J. B., R. C. Aster, and P. R. Kyle (2004), Volcanic eruptions observed with infrasound, Geophys. Res. Lett., 31, L14604,

doi:10.1029/2004GL020020.Kim, K., J. M. Lees, and M. Ruiz (2012), Acoustic multipole source model for volcanic explosions and inversion for source parameters,

Geophys. J. Int., 191(3), 1192–1204, doi:10.1111/j.1365-246X.2012.05696.x.Kim, K., D. Fee, A. Yokoo, and J. M. Lees (2015), Acoustic source inversion to estimate volume flux from volcanic explosions, Geophys. Res.

Lett., 42(13), 5243–5249, doi:10.1002/2015GL064466.Lacanna, G., M. Ichihara, M. Iwakuni, M. Takeo, M. Iguchi, and M. Ripepe (2014), Influence of atmospheric structure and topography on

infrasonic wave propagation, J. Geophys. Res. Solid Earth, 119, 2988–3005, doi:10.1002/2013JB010827.Lamb, O. D., S. D. Angelis, and Y. Lavallée (2015), Using infrasound to constrain ash plume rise, J. Appl. Volcanol., 4(1),

doi:10.1186/s13617-015-0038-6.Lavallée, Y., et al. (2015), Thermal vesiculation during volcanic eruptions, Nature, 528(7583), 544–547.

AcknowledgmentsThe authors acknowledge supportfrom the Instituto Nacional deSismología, Vulcanología, Meteorologíae Hidrología (INSIVUMEH), Guatemala,and the staff of the Santiaguito VolcanoObservatory (OVSAN). We are gratefulto our local guide and networkmanager in Guatemala, A. Pineda. Wethank K. Kim for advice on the acousticmultipole source inversion and sharingsome of the scripts used in this study.Y. Lavallée acknowledges support fromthe European Research CouncilStarting Grant on Strain Localisationin Magma (SLiM, 306488). S. DeAngelis and A. Rietbrock are supportedby the Liverpool Earth Observatory,University of Liverpool. The data usedin this study can be requested fromthe authors.

DE ANGELIS ET AL. EXPLOSIONS AT SANTIAGUITO 6226

Geophysical Research Letters 10.1002/2016GL069098

Marchetti, E., M. Ripepe, A. Harris, and D. D. Donne (2009), Tracing the differences between Vulcanian and Strombolian explosions usinginfrasonic and thermal radiation energy, Earth Planet. Sci. Lett., 279(3–4), 273–281, doi:10.1016/j.epsl.2009.01.004.

Matoza, R. S., D. Fee, M. A. Garcés, J. M. Seiner, P. A. Ramón, and M. A. H. Hedlin (2009), Infrasonic jet noise from volcanic eruptions,Geophys. Res. Lett., 36, L08303, doi:10.1029/2008GL036486.

Patrick, M. R. (2007), Dynamics of Strombolian ash plumes from thermal video: Motion, morphology, and air entrainment, J. Geophys. Res.,112(B6), doi:10.1029/2006JB004387.

Ripepe, M., C. Bonadonna, A. Folch, D. D. Donne, G. Lacanna, E. Marchetti, and A. Höskuldsson (2013), Ash-plume dynamics and eruptionsource parameters by infrasound and thermal imagery: The 2010 Eyjafjallajökull eruption, Earth Planet. Sci. Lett., 366, 112–121,doi:10.1016/j.epsl.2013.02.005.

Rossing, T. (2016), Springer Handbook of Acoustics, Springer-Verlag, New York.Valade, S., A. Harris, and M. Cerminara (2014), Plume Ascent Tracker: Interactive Matlab software for analysis of ascending plumes in image

data, Comput. Geosci., 66, 132–144, doi:10.1016/j.cageo.2013.12.015.Vergniolle, S., and J. Caplan-Auerbach (2004), Acoustic measurements of the 1999 basaltic eruption of Shishaldin Volcano, Alaska,

J. Volcanol. Geotherm. Res., 137(1–3), 135–151, doi:10.1016/j.jvolgeores.2004.05.004.von Aulock, F., Y. Lavallée, A. Hornby, O. Lamb, B. Andrews, and J. Kendrick (2016), Observing changes at Santiaguito Volcano, Guatemala

with an Unmanned Aerial Vehicle (UAV), EGU General Assembly Abstracts, 18, EGU2016–11080.Yamamoto, H., I. M. Watson, J. C. Phillips, and G. J. Bluth (2008), Rise dynamics and relative ash distribution in vulcanian eruption

plumes at Santiaguito Volcano, Guatemala, revealed using an ultraviolet imaging camera, Geophys. Res. Lett., 35, L08314,doi:10.1029/2007GL032008.

DE ANGELIS ET AL. EXPLOSIONS AT SANTIAGUITO 6227